NVIDIA forecasts first $100bn quarter as Data Center revenue surges 117%

NVIDIA Corporation delivered another extraordinary acceleration in fiscal second-quarter growth as revenue reached $96.2 billion, up 106% year over year and 18% from the previous quarter, while Data Center revenue surged 117% to $89.0 billion. GAAP operating income more than doubled to $63.7 billion and net income climbed 126% to $59.7 billion, producing diluted earnings of $2.46 per share. Non-GAAP earnings reached $2.22 per share, above the approximately $2.08 consensus going into the report, while revenue also exceeded Wall Street expectations of roughly $91.9 billion. Management expects Q3 revenue of approximately $108 billion, which would mark NVIDIA’s first quarter above $100 billion and stands above the $104.19 billion analyst consensus compiled by LSEG.

The results confirm that the global AI infrastructure buildout is still expanding at a pace rarely seen at NVIDIA’s scale. Data Center now accounts for more than nine-tenths of total revenue, while the company’s next-generation Vera Rubin platform has moved into full production alongside continued Blackwell demand. NVIDIA is also broadening its role beyond GPUs through CPUs, networking, AI inference accelerators, financing partnerships and physical infrastructure associated with the construction of large AI factories.

The market reaction was more restrained. NVIDIA shares slipped roughly 1.2% in extended trading despite the earnings beat and stronger-than-expected guidance, highlighting how much future growth is already embedded in investor expectations. The debate is increasingly shifting away from whether AI spending is growing and toward whether NVIDIA can retain its current share of that spending as hyperscalers develop more custom silicon and rivals target inference workloads.

Data Center revenue jumps 117% as AI infrastructure spending remains the dominant growth engine

NVIDIA’s Data Center business generated $89.0 billion during Q2, up 18% sequentially and 117% from a year earlier. That segment alone is now nearly twice the size of NVIDIA’s entire company revenue from the comparable quarter last year, when total sales were $46.7 billion.

The growth reflects continued capital spending by cloud providers, AI laboratories, technology companies and governments building increasingly large computing clusters. Microsoft and Meta are among the companies planning aggressive AI infrastructure investment, while industry-wide spending by major technology groups is expected to exceed $730 billion this year compared with roughly $400 billion last year.

That spending cycle is creating demand across NVIDIA’s full Data Center stack rather than GPUs alone. Vera Rubin racks are now running at partners including CoreWeave, Google Cloud, Microsoft Azure, Oracle Cloud Infrastructure and Nebius, while NVIDIA is also rolling out new networking systems and the Vera CPU designed specifically for AI-agent workloads.

The transition from Blackwell toward Vera Rubin is strategically important because NVIDIA must maintain rapid product cycles without creating prolonged pauses in customer purchasing. Large customers can delay orders when a major new architecture is close to launch, so successfully ramping a new generation while maintaining strong demand for existing products is one of the main operational challenges in the AI accelerator market.

NVIDIA’s Q2 performance suggests that transition is currently working. Revenue increased 18% sequentially despite the scale of the prior quarter, showing that customers are continuing to deploy infrastructure rather than waiting for future systems.

The company is also pushing deeper into AI inference, where workloads involve running trained models rather than training them. That market is expected to become increasingly important as enterprises and consumers make greater use of AI agents, search tools, coding systems and other applications that continuously generate tokens.

$108 billion Q3 forecast suggests AI spending is still accelerating despite China exclusion

NVIDIA expects Q3 revenue of approximately $108 billion, plus or minus 2%. The midpoint exceeds the roughly $104.19 billion analyst consensus and would represent another sequential increase from Q2’s $96.2 billion.

The guidance is particularly notable because NVIDIA is assuming no Data Center compute revenue from China. That makes the forecast a cleaner indication of demand elsewhere and suggests the U.S., Europe, Middle East and other Asian markets are currently sufficient to support continued expansion without relying on Chinese accelerator sales.

China nevertheless remains strategically important. Export restrictions have limited NVIDIA’s ability to sell its most advanced processors into one of the world’s largest technology markets, while domestic Chinese companies are developing alternative AI chips and software ecosystems.

That creates a long-term competitive risk even if the immediate financial impact is manageable. Restrictions can accelerate local substitution and encourage customers to invest in non-NVIDIA architectures, potentially reducing the company’s influence in a major market over time.

For now, the geographic breadth of AI investment appears to be compensating. NVIDIA has described sovereign AI infrastructure, new frontier laboratories, startups and physical AI applications as additional demand drivers alongside established hyperscalers.

The Q3 forecast also reinforces the extraordinary speed of NVIDIA’s revenue transformation. The company generated $46.7 billion in Q2 of the prior fiscal year and is now guiding toward more than twice that amount in a single quarter only one year later.

Operating income rises 124% as NVIDIA preserves a 75% gross margin at massive scale

GAAP gross margin reached 75.0% during Q2 compared with 72.4% a year earlier. Non-GAAP gross margin was also 75.0%, showing that NVIDIA has been able to maintain premium pricing and product economics even as quarterly revenue approaches $100 billion.

Operating income increased 124% to $63.7 billion, substantially faster than the 106% revenue increase. Operating expenses rose 55% to $8.4 billion, meaning the incremental cost required to support growth remained far below the increase in gross profit.

That operating leverage remains one of the most remarkable aspects of NVIDIA’s financial model. A large semiconductor company typically faces significant manufacturing, R&D and sales costs as revenue expands, but NVIDIA’s outsourced manufacturing structure and software-driven ecosystem allow a substantial portion of incremental gross profit to reach operating income.

Research and development spending still reached more than $7 billion in the quarter, reflecting the enormous cost of maintaining annual architecture cycles across GPUs, CPUs, networking, software and robotics. Even so, the expense remains modest relative to the revenue generated by the products those investments support.

Management expects Q3 gross margin of approximately 74%, plus or minus 50 basis points. A modest decline from Q2 would still leave profitability at historically exceptional levels while Vera Rubin ramps and the product mix changes.

Margin durability will remain closely watched because competition is expanding. AMD and Intel are targeting AI inference, hyperscalers are developing proprietary processors, and several specialized accelerator companies are attempting to build alternatives for particular workloads.

NVIDIA expands beyond GPUs as Vera Rubin, Groq and AI financing reshape the competitive moat

NVIDIA is increasingly positioning itself as the operating platform for AI infrastructure rather than merely a semiconductor supplier. Vera Rubin combines compute, networking and software across full-scale AI factory deployments, while the company is simultaneously expanding into CPUs and inference acceleration.

The Groq relationship is particularly important in inference. NVIDIA has incorporated technology licensed from Groq through a roughly $17 billion transaction and is pairing those capabilities with Vera Rubin as demand shifts toward high-volume, low-latency AI responses.

That strategy addresses one of the main threats to NVIDIA’s current position. Custom chips can sometimes perform inference workloads more efficiently than general-purpose GPUs, especially when models and use cases are stable enough to justify specialized hardware.

NVIDIA is attempting to defend its position by offering customers a broader computing architecture rather than competing processor by processor. CUDA software, networking, system design, CPUs and AI accelerators can make the full platform harder to replace even when another vendor offers a competitive chip.

The company is also becoming more involved in financing AI infrastructure. NVIDIA announced partnerships with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR aimed at mobilizing more than $500 billion of third-party capital over time for AI infrastructure development.

That move can accelerate demand by helping customers finance extremely capital-intensive data centers, but it also creates a new area of scrutiny. Investors will increasingly watch whether the AI ecosystem is generating sufficient economic returns independently or becoming reliant on increasingly complex financing arrangements between chip suppliers, infrastructure developers and financial institutions.

Free cash flow remains enormous even as working capital and infrastructure investment increase

NVIDIA generated $24.1 billion of operating cash flow during Q2 compared with $15.4 billion a year earlier. After capital expenditures and related payments, free cash flow reached approximately $21.3 billion, up from $13.5 billion in the comparable period.

Free cash flow was lower than the exceptional $48.6 billion produced in Q1 because working-capital requirements expanded sharply. Accounts receivable increased significantly as revenue grew, while inventory also rose as NVIDIA prepared for continued customer demand and product ramps.

Inventory reached $31.6 billion at the end of July compared with $21.4 billion in January. Accounts receivable increased to $63.1 billion from $38.5 billion over the same period, reflecting the enormous increase in business volume moving through NVIDIA’s supply chain.

Those increases are not inherently negative if they support future growth, but they add another operational dimension to the AI boom. NVIDIA must coordinate foundry capacity, advanced packaging, memory, networking and system assembly across a rapidly expanding ecosystem without overbuilding inventory if customer demand changes.

The company’s overall financial position remains exceptionally strong. Cash and cash equivalents reached $22.4 billion, while marketable debt and equity securities add substantial additional liquidity. Total shareholder equity increased to approximately $229 billion from $157 billion at the beginning of the fiscal year.

That financial capacity allows NVIDIA to simultaneously fund R&D, infrastructure partnerships, acquisitions and shareholder returns without placing meaningful pressure on liquidity.

$26 billion shareholder return shows how quickly AI profits are being converted into capital distributions

NVIDIA returned approximately $26 billion to shareholders during Q2 through share repurchases and dividends. The company still had approximately $99 billion remaining under its repurchase authorization at quarter-end.

Repurchase payments totaled roughly $19.7 billion during the quarter, while dividends paid reached approximately $6.0 billion. NVIDIA will pay another quarterly dividend of $0.25 per share on October 1 to shareholders of record on September 10.

The scale of the capital-return program underscores how rapidly NVIDIA’s cash-generation profile has changed. The company can now spend tens of billions of dollars on buybacks while simultaneously increasing investment across hardware, software and AI infrastructure.

The investment case is nevertheless less about shareholder distributions than the durability of future AI spending. NVIDIA’s valuation already reflects an expectation that artificial intelligence will become a core layer of global computing infrastructure rather than a temporary capital-spending cycle.

Reuters reported that technology companies are increasingly directing some AI spending toward internally developed processors as they attempt to reduce dependence on NVIDIA’s expensive and capacity-constrained hardware. That trend represents one of the most important long-term risks to the company’s current market share.

The counterargument is that the overall AI infrastructure market could expand quickly enough for NVIDIA to continue growing even as competitors take some share. Management believes the revenue opportunity for NVIDIA AI chips could exceed $1 trillion through 2027, compared with the $500 billion opportunity previously cited through 2026.

The muted after-hours stock reaction captures that debate. Revenue growth above 100%, a stronger Q3 forecast and 75% gross margins would represent extraordinary results for almost any company, but NVIDIA is being valued against an expectation of continued technological and financial dominance.

The next several quarters will show whether Vera Rubin, inference growth and broader AI infrastructure deployment can sustain that dominance as competitors increase spending on their own processors. If Data Center growth remains above the broader industry and gross margins remain in the mid-70% range, NVIDIA’s current growth cycle could continue far longer than conventional semiconductor cycles.

Key takeaways from NVIDIA’s $96 billion quarter and $108 billion Q3 revenue forecast

• Q2 revenue surged 106% to $96.2 billion, demonstrating that NVIDIA’s AI infrastructure growth is still accelerating despite its already enormous revenue base.
• Data Center revenue reached $89.0 billion, up 117%, making AI computing the overwhelming driver of NVIDIA’s financial performance and strategic value.
• GAAP operating income increased 124% to $63.7 billion as revenue growth continued to outpace operating expenses and preserve exceptional operating leverage.
• Gross margin held at 75.0%, showing NVIDIA continues to maintain premium economics even as it scales quarterly revenue toward the $100 billion threshold.
• Q3 revenue guidance of $108 billion exceeded the roughly $104.2 billion consensus and assumes no Data Center compute revenue from China.
• Vera Rubin has entered full production, making its ramp the next major test of NVIDIA’s ability to sustain annual product cycles without disrupting customer spending.
• Free cash flow reached $21.3 billion despite substantial working-capital investment, leaving NVIDIA with extraordinary capacity to fund R&D, infrastructure and capital returns.
• NVIDIA returned approximately $26 billion to shareholders during Q2 and still has about $99 billion remaining under its repurchase authorization.
• Competition is increasingly shifting toward inference and custom chips, making hyperscaler in-house silicon one of the largest long-term risks to NVIDIA’s current dominance.
• Shares slipped roughly 1% after hours despite the beat and higher outlook, showing investors now expect NVIDIA to deliver extraordinary growth rather than merely strong growth.